Read only flag set automatically, data inconsistency between master/replica.

[prisms....@gmail.com]

Today morning, my SQL instances and replica both were automatically set to read_only ON.

I set the flags manually to off.

Further, I found that there is data inconsistency between replica and master. How do I find out if the connection is broken? How do I restore/sync the master to the replica?

I also created a test instance with a replica and found that read_only was set to ON though the default should be off. Right?

[prisms....@gmail.com]

This is for 2nd generation instances

[prisms....@gmail.com]

Ok, looks like the read only flags are working fine.
But the master keeps failing randomly and when it goes to failover, I get the The MySQL server is running with the --read-only

According to the docs, when the master fails the replica becomes the new master. Any way to trigger this manually, or figure out why the master is failing?

[Adam (Cloud Platform Support)]

My advice is to file a production issue on https://code.google.com/p/googlecloudsql/issues/list, and provide as much detail as possible (you can make the issue private) so that we can look into it.